High Performance Liquid Chromatography Method for the Determination of Pyridoxal-5-Phosphate in Human Plasma

Home , Pyridoxal phosphate

High performance liquid chromatography method for the determination of pyridoxal-5-phosphate in human plasma: How appropriate are cut-off values for vitamin B6 de®ciency?

AL Bailey1*, AJA Wright1 and S Southon1

1Nutrition, Diet and Health Department, Institute of Food Research, Norwich Research Park, NR4 7UA, UK.

Objectives: Application of a HPLC (high performance liquid chromatography) method, using cyanide derivatisation, to the determination of plasma pyridoxal-5-phosphate (PLP) concentrations as an indicator of vitamin B6 adequacy. Setting: The study was performed at the Institute of Food Research, Norwich, UK. Blood samples were taken at the Institute, at Health Centres, or in the volunteer's home. Subjects: 51 adolescent, 131 adult, 68 non-institutionalized elderly and 44 aged (>73 y) volunteers were recruited from local authority schools, local Health Centres and General Practitioners. Results: The mean PLP recovery was 92.8%. The intra- and inter-assay coef®cients of variation were 2.8% and 5.2% respectively. Mean PLP concentrations for males and females, respectively, were: adolescents (13 ± 14 y), 36.4 and 43.5 nM; adults (20 ± 64 y), 39.2 and 40.0 nM; elderly (68 ± 73 y), 34.8 and 35.3 nM; aged (>73 y), 57.8 and 49.0 nM. Percentages of subjects with PLP concentrations < 34.4 nM were over 26% in all population groups. Mean vitamin B6 intakes (mg=g protein intake), as assessed by weighed dietary records, were all above reference nutrient intakes (15 mg=g protein). Conclusions: An HPLC method, using cyanide derivitisation, has been applied to the determination of plasma PLP. Comparisons of results for local population groups with current cut-off values for plasma PLP, show large numbers of volunteers at risk of vitamin B6 de®ciency although this is not re¯ected by vitamin B6 intakes calculated from food tables. The 34.4 nM cut-off value for value for plasma PLP, indicating de®ciency, is questioned. Sponsorship: UK Ministry of Agriculture, Fisheries and Food; Department of Education and Science; Vitamin Forum. Descriptors: Vitamin B6; Plasma pyridoxal-5-phosphate; HPLC; Cyanide derivative

Introduction The total body pool of vitamin B6 is around 200 mg (Coburn et al, 1988) (Note: not 20 ± 30 mg as quoted by Vitamin B (active 3-hydroxy-2-methylpyridine deriva- 6 Gibson, 1990). About 70 ± 80% of this pool is in skeletal tives) is an essential precursor for the pyridoxal-5-phos- muscle and is resistant to depletion, reducing only after 6 phate (PLP) coenzyme for protein and, to a lesser extent, weeks of low intake (1.76 mmol=d; 0.37 mg=d pyridoxine carbohydrate and lipid metabolism. More recently a range hydrochloride) in adult males (Coburn et al, 1991). How- of additional roles have been suggested for this vitamin: (i) ever, the major transport form of vitamin B is PLP, which prevention of atherosclerorosis (Murray & Pizzorno, 6 declines relatively rapidly at low dietary intake. Thus, 1990a; Vermaak et al, 1987; Ubbink, 1994; Verhoef & measurement of this active coenzyme form of vitamin B Stampfer, 1995; Selhub et al, 1993); (ii) optimization of 6 in plasma is more likely to re¯ect short-term intake than immune function (Rall & Meydani, 1993); (iii) treatment of muscle B which is tightly bound to glycogen phosphoryl- premenstrual tension (PMT) (Murray & Pizzorno, 1990b; 6 ase and only turns over as this enzyme does (Coburn et al, Kleijnen, et al, 1991); (iv) prevention of nausea and 1991). Low plasma concentrations of PLP, therefore, do not vomiting during pregnancy (Vir et al, 1980; Guilarte, re¯ect muscle concentration but may provide a good 1993; (v) learning and memory (Guilarte, 1993; Kleijnen indication of de®ciency risk. & Knipschild, 1991); (vi) alteration of steroid hormone Biochemical methods, developed over the last twenty responsiveness, and thus, the aetiology and treatment of years for the determination of PLP concentrations in certain cancers and hypertension (Bender, 1994); (vii) biological samples, include microbiological, spectrophoto- treatment of psychiatric patients (Sandyk & Pardeshi, metric, ¯uorometric, enzymatic, radio-enzymatic, and chro- 1990; Bender, 1992; Stewart et al, 1984). matography (Sauberlich, 1984, Vanderslice et al, 1985; Leklem, 1990). No single method has been recommended for determination of vitamin B6 adequacy (Leklem, 1990; *Correspondence (and guarantor): AL Bailey, Nutrition, Diet and Health Li & Lumeng, 1980; Bender, 1993). Department, Institute of Food Research, Norwich Research Park, NR4 Separation of PLP from other B vitamins is easily 7UA, UK. 6 Received 7 November 1998; revised 14 January 1999; accepted 16 January achieved by high performance liquid chromatography 1999. (HPLC), but detection has to be achieved by formation of Plasma PLP by HPLC AL Bailey et al 449 UV absorbing, ¯uorescent or electrochemically active deri- 64 y, n 131, 64 males, 67 females) were recruited by vatives, since PLP has only low intrinsic ¯uorescence. The written invitation, at random from the General Practi- dif®culty with HPLC methods for the determination of tioners' lists of two local Health Centres (Norwich and plasma PLP has been their lack of sensitivity, since Dereham). Non-institutionalized elderly volunteers (68 ± 73 plasma PLP concentrations are typically in the 5 ± years, n 68 29 males, 39 females) and aged volunteers 100 ng=ml (20.3 ± 405 nM) range, necessitating relatively (74 ± 90 y, n 44, 15 males, 29 females) were randomly large sample sizes (0.3 ± 1.0 ml) (Sauberlich, 1984). selected from the age=gender registers of two General The use of cyanide to produce a highly ¯uorescent Practitioners in Norwich and recruited by written invita- derivative of PLP was ®rst used in 1959 (Ohishi & tion. Recruitment methods are described in full detail by Fukui, 1968, Adams, 1969, Tamura & Takanaski, 1970) Wright et al, (1995). and enabled more sensitivity than other derivatives. Cya- nide derivatisation of PLP was used in combination with HPLC in methods by Naoi et al (1988), who determined Sample preparation PLP concentrations in puri®ed enzymes, human brain Fresh, fasting (12 h) venous, whole blood was removed by homogenates and tissue cultured cells, but not human venepuncture into heparinized plastic tubes (Sarstedt, Lei- plasma. cester, UK), centrifuged (5006g,4C, 10 min) and 500 ml In the present paper, a method based on that of Naoi et of plasma removed into 2 ml plastic microfuge tubes al (1988) was used to measure PLP concentrations in small (Sarstedt, Leicester, UK) which were stored at 40 C (100 ml) volumes of plasma. This method facilitates studies until analysis. where small samples may be important, for example studies involving very young volunteers, or when a single sample Derivatisation needs to be used for a variety of biochemical assays. Minor Derivatisation of samples was performed under non-UV modi®cations to the method are described and reproduci- lighting. Duplicate aliquots (100 ml) of thawed plasma were bility and recovery data are presented, together with ranges pipetted into further microfuge tubes containing aqueous of PLP concentrations found in human adolscent adult, potassium phosphate buffer (100 ml, 10 mM; pH 7.4) and elderly and aged population groups. The incidence of deproteinized by the addition of trichloroacetic acid (TCA; vitamin B6 de®ciency, as judged by currently used cut-off 200 ml, 10% w=v). The samples were capped and heated for value for plasma PLP, was determined. How appropriate 15 min at 50C. After cooling to room temperature (20 ± these cut-off values are, as indicators of de®ciency, is 25C), dipotassium phosphate (140 ml, 3.3 M; pH 10.2) and considered. potassium cyanide (40 ml, 8 mM) were added, mixing tube contents after each addition. The cyanide derivatives of Materials and methods PLP and PA were formed by heating again at 50C for 25 min. After again cooling to room temperature the Reagents ¯uorescent derivatives were stabilized by addition of phos- All reagents were analytical grade unless speci®ed other- phoric acid (50 ml, 28%). Samples were centrifuged at wise. Pyridoxal-5-phosphate (PLP) and 4-pyridoxic acid 88006g for 5 min and the supernatants transferred to (PA) were obtained from Sigma Chemical Co, Dorset, UK. amber, crimp cap, auto sampler vials for HPLC analysis. All water was double glass distilled. Potassium cyanide was obtained from Sigma Chemical Co. (Dorset, UK). Pyridox- ine (PN), Pyridoxal (PL), pyridoxamine (PM) and pyridox- Chromatographic conditions amine-5-phosphate were also obtained from Sigma The analytical HPLC column was an ApexTM 3 mm ODS Chemical Company. (C18) I, 2560.4 cm id. (Jones Chromatography, Mid Gla- morgan, UK) instead of that used by Naoi et al (1988), Pyridoxal-5-phosphate and 4-pyridoxic acid standards (STRTM ODS-H; 150 mm64 mm id; 5 mm particle size; A stock PLP solution (50 mg=ml; 202.3 mM in 10 mM 100 AÊ pore size). Samples (50 ml) were injected by loop potassium phosphate buffer, pH 7.4) was prepared freshly ®lling using a RheodyneTM 7125 valve (Anachem, Luton, each week and stored at 4C in the dark. Working Bedfordshire, UK). The analytical column was protected by standard solutions were prepared daily by dilution of the use of a guard column packed with 40 mm C18 material stock 25 mg=l (0.1 mM) in 10 mM phosphate buffer. 2cm62 mm id, Anachem, Luton, UK). The HPLC pump For the additional determination of pyridoxic acid (PA), used was isocratic (SA 6410B, Severn Analytical, Hitchin a stock solution of PA (100 mg=l; 0.546 mM) was prepared Beds, UK). The mobile phase was 1 mM heptane sulphonic in 10 mM phosphate buffer and diluted together with PLP acid in 2 M potassium acetate, adjusted to pH 3.75 with to give a mixed PLP and PA standard (25 mg=l PLP and potassium hydroxide (solid). The ¯ow rate was 1.0 ml=min. 50 mg=l PA). For the purpose of peak identi®cation PL, PM, The column temperature was ambient. Following separa- PN and PMP were prepared individually and mixed with tion, cyanide derivatives were detected ¯uorometrically PLP and PA at ®nal working concentrations of 25 mg=l. using a LS 5 luminescence spectrophotometer (Perkin Elmer, Buckinghamshire, UK). Excitation and emission Human volunteers and ethics wavelengths were 318 and 418 nm respectively. Concen- The study was approved by the Institute of Food Research's trations of PLP or PA were calculated by comparison of Ethics Committee prior to recruitment of volunteers. All peak heights with those of standard solutions following volunteers consented to have blood samples taken by a collection of data using APEX Chromatography data col- fully quali®ed phlebotomist and record food intake after lection system (Severn Analytical, Hitchin, Beds, UK). A full explanation of study and any risks involved. Adoles- KromasilTM 5 mm ODS I (15 cm60.4 mm id; BAS Techni- cents (13 ± 14 years, n 51, 18 Males, 33 Females) were col, Stockport, UK) analytical column was also tested as an recruited from two local authority schools. Adults (20 ± alternative to the Apex column mentioned above. Plasma PLP by HPLC AL Bailey et al 450 Sample stability and quality control Recovery of PLP from plasma samples was determined by analysing plasma with and without the addition of working standard before derivatization, to give a spike PLP concentration of 12.5 mg=l plasma (50.6 nM). Recoveries were calculated as (spiked value minus unspiked value) as a percentage of the unspiked value. The possibility of establishing a quality control system for routine daily use was investigated. Pooled, fresh, heparinized, whole blood (total volume 100 ml) was centrifuged (10 min; 500 g force), the plasma removed, pooled and mixed gently to ensure homogeneity. Aliquots (100 ml) of this pooled plasma were stored in plastic microfuge tubes (2 ml, Sarstedt, Leicester UK), under liquid nitrogen and removed at intervals of up to 16 months for PLP analysis. These QC samples were also used to measure between assay coef®cient of variation.

Vitamin B6 intakes Vitamin B6 intakes for the populations described in this study were measured by 5 ± 7d weighed dietary records and calculated from food tables. Detailed description of the methods used are described elsewhere (Wright et al, 1995; Southon et al, 1992). Vitamin B6 intakes were expressed as mg=mg of protein consumed and compared to reference nutrient intakes (RNI), estimated average requirements (EAR) and lower reference nutrient intakes (LRNI) for the relevant population age group (Department of Health, 1991). Percentages of each population with vitamin B6 intakes below each of these cut-offs were calculated.

Statistical methods Linearity of PLP calibration curves was tested by regres- Figure 1 HPLC chromatogram showing separation of pyridoxal-5-phos- sion analysis. The distribution of values for both plasma phate (PLP), 4-pyridoxic acid (PA), pyridoxamine (PM) and pyridox- PLP concentration (nM) and vitamin B6 intake (mg=g amine-5-phosphate (PMP) using an Apex I 3 mm ODS analytical column (25 cm60.4 cm id). All other chromatographic conditions are described in protein intake) were normalized by mathematical (log10) transformation prior to further statistical investigation. the methods section. Mean (log10) PLP concentrations and vitamin B6 intakes were compared, within gender, by one-way analysis of analytical column used by Naoi et al, (1988). Both the variance (1-ANOVA). Following a signi®cant (P < 0.05) ApexTM or a tromasilTM ODS I were found to adequately variance ratio (F-test), any differences between mean separate the forms of vitamin B6 with no noticeable values for the four age groupings (adolescent, adult, elderly differences in resolution. and aged) were determined by the post-hoc application of the ScheffeÂ test. Female values were compared to those of their male counterparts by Student's unpaired t-test. Within gender; relationships, if any, between plasma PLP concentrations (log10) and vitamin B6 intakes (log10) were Linearity, recovery, variability and detection limit assessed using linear regression analysis. Calibration curves (0, 5, 10, 15, 20, 25 ng=ml) for PLP performed daily were linear. Regression analysis of 10 calibrations, spanning a period of 1 month, indicated a Results mean degree of association (r2) of 0.9985 (s.d. 0.0005; Chromatographic separation range 0.9980 to 0.9990); mean intercept of 0.0616 mV The chromatographic separation of pyridoxal-5-phosphate (s.d. 0.5211, range 1.1176 to 0.7855) and a mean slope of (PLP), pyridoxic acid (PA), pyridoxamine (PM) and pyr- 3.1712 mV=ng.ml 1 (s.d. 0.2530; range 2.9275 to 3.6681). idoxamine-5-phosphate (PMP), as their cyanide deriva- The mean recovery of added PLP from plasma was tives, is shown in Figure 1. Pyridoxal (PL) and 92.8% (n 10, CV 2.03%; range 89.7% ± 96.6%). The pyridoxine (PN) did not produce a ¯uorescence response within assay coef®cient of variation for plasma was 2.8% at concentrations of 25 mg=l. Both PM and PMP gave a (n 10, mean 11.19 mg=l (45.32 nM)). The between assay ¯uorescence response, but not at suf®cient sensitivity to coef®cient of variation (CV), calculated from analysis of allow determination of these forms of vitamin B6 in plasma QC samples over a three month period was 5.2% (n 17, samples. A typical chromatogram of a derivatised plasma mean 13.84 mg=l (56.05 nM)). Similarly at higher PLP sample is shown in Figure 2. concentrations over 9 months CV 5.7% (n 5, mean- The data given in this paper were achieved using an 23.82 mg=l (96.47 nM). The detection limit for plasma ApexTM I column (Jones Chromatography, Mid Glamor- PLP, calculated as three times chromatographic base line gan, UK) instead of the STR ODS H (Shimadzu, Japan) noise, was 2 mg=l (8 nM). Plasma PLP by HPLC AL Bailey et al 451 3, 9 and 16 months, indicated no signi®cant loss of PLP with time.

Plasma PLP concentrations in human volunteers Log10 mean and standard deviation (s.d.), together with geometric mean and range (min ± max), of PLP concentrations found in adolescent, adult, elderly and aged volunteers, and the percentage of subjects with plasma PLP levels indicative of inadequate vitamin B6 status (<34.4 nM; Rose et al, 1976), are shown in Table 1. Aged males (>73 y) had signi®cantly (P < 0.05) higher PLP concentrations than all their younger counterparts, between which there were no signi®cant differences. There were no differences between the female age groups, or between males and females at any age.

Vitamin B6 intakes Log10 mean and s.d., together with geometric mean and range (min ± max) of vitamin B6 intakes (mg=d) for adolescent, adult, elderly and aged populations are shown in Table 2. The same data, expressed as mg=g protein intake, are shown in Table 3. Percentages of each population with vitamin B6 intakes below reference nutrient intakes (RNI; 1.2 mg=d for adolescent males; 1.0 mg=d for adolescent females; 1.4 mg=d for adult, elderly and aged males; 1.2 mg=day for adult, elderly and aged females; 15 mg=g protein intake for all populations), estimated average requirements (EAR; 13 mg=g protein intake) and lower reference nutrient intakes (LRNI; 11 mg=g protein intake) (Department of Health, 1991) are also shown in Tables 2 and 3 as appropriate. Figure 2 HPLC chromatogram of a derivatised plasma sample, showing When expressed as total vitamin B6 intake (mg=day) pyridoxal-5-phosphate (PLP) and pyridoxic acid (PA). Chromatographic adult males had signi®cantly higher (P < 0.001) intakes conditions are described in the methods section. than younger or older male volunteers, and sign®cantly higher (P < 0.05) intakes than adult females. Aged females had signi®cantly lower (P < 0.001) vitamin B6 intakes than Sample stability adult females, and lower (P < 0.05) than their male coun- Linear regression analysis of data from 3 QC batches of terparts. When expressed as mg=g protein intake, there were human plasma, stored under liquid nitrogen ( 196 C) for no differences between the male age groups in vitamin B6

Table 1 Plasma pyridoxal-5-phosphate concentrations (nmol=l) found in adolescent, adult, elderly and aged volunteers{{.

1-way Analysis of Variance Adolescents Adults Elderly Aged Signi®cance of Variance (13 ± 14 y) (20 ± 64 y) (68 ± 73 y) (74 ± 90 y) Ratio (F-test)

Males (n) 18642915 Mean s.d. 1.561a 0.136 1.593a 0.184 1.541a 0.246 1.762b 0.232 P < 0.01 (log 10) GM{ (nmol=l) 36.39 39.17 34.75 57.81 (range; min-max) (23.49 ± 77.76) (14.99 ± 85.05) (14.70 ± 165.10) (25.03 ± 127.82) % at risk of 50.0% 37.5% 48.3% 26.7% de®ciencyx Females (n) 33 67 39 29 Mean s.d. 1.638x0.157 1.602x0.243 1.548x0.215 1.690x0.207 P < 0.05{ (log 10) GM{ (nmol=l) 43.45 40.00 35.32 48.98 (range; min-max) (25.11 ± 105.71) (13.37 ± 387.99) (15.30 ± 268.60) (21.10 ± 132.88) % at risk of 27.3% 46.3% 59.0% 27.6% de®ciencyx

x: Volunteers with plasma PLP concentrations <34:4 nmol=l are judged as being at risk of vitamin B6 de®ciency (Rose et al, 1976), but we suggest that this cut-off was placed arbitrarily. {Geometric mean; i.e. antilog 10 of log 10-mean. abc (xyz): Within males (abc) or females (xyz), means not sharing a common superscript are signi®cantly (P < 0.05) different; Post-Hoc ScheffeÂ test. {Though Variance Ratio is signi®cant, the Post-Hoc ScheffeÂ test indicates no difference between individual means. Note: Female means are not signi®cantly different (P < 0.05) from their corresponding male means; Student's unpaired t-test. {{Means and s.d. of log 10-transformed data, geometric mean, ranges (min-max) and apparent percentage of subjects at risk of vitamin B6 de®ciencyx. Plasma PLP by HPLC AL Bailey et al 452

Table 2 Vitamin B6 intakes (mg=d) for adolescent, adult, elderly and aged volunteers{{. 1-way Analysis of Variance Adolescents Adults Elderly Aged Signi®cance of Variance (13 ± 14 y) (20 ± 64 y) (68 ± 73 y) (74 ± 90 y) Ratio (F-test)

Males (n) 18 64 29 15 Mean s.d. 0.076a 0.107 0.195b 0.109 0.093a 0.115 0.102a 0.105 P < 0.001 (log 10) GM{ (mg=d) 1.19 1.57 1.24 1.26 (range; min-max) (0.75 ± 1.87) (0.85 ± 2.66) (0.71 ± 2.04) (0.85 ± 1.82) % < RNIx 44.4% 26.6% 62.1% 60.0% Females (n) 33 67 39 29 Mean s.d. 0.038xy0.081 0.095y 0.139* 0.052xy 0.122 0.023x 0.143* P < 0.001 (log 10) GM{ (mg=d) 1.09 1.24 1.13 0.95 (range; min-max) (0.66 ± 1.56) (0.45 ± 2.43) (0.63 ± 2.00) (0.52 ± 1.52) % < RNIx 27.3% 41.8% 51.3% 69.0%

x: Reference nutrients intake (RNI): adolescent males 1.2 mg=d; adolescent females 1.0 mg=d; adult, elderly and aged males 1.4 mg=d; adult, elderly and aged females 1.2 mg=d; (Department of Health, 1991). {Geometric mean; i.e. antilog 10 of log 10-mean. abc (xyz): Within males (abc) or females (xyz), means not sharing a common superscript are signi®cantly (P < 0.05) different; Post-Hoc ScheffeÂ test. Female mean is signi®cantly different (P < 0.05) from its corresponding male mean; Student's unpaired t-test. {{Means and s.d. of log 10-transformed data, geometric mean, ranges (min-max) and percentage of subjects with intakes below RNIx

Table 3 Vitamin B6 intakes (mg=g protein intake) for adolescent, adult, elderly and aged volunteers{{. 1-way Analysis of Variance Adolescents Adults Elderly Aged Signi®cance of Variance (13 ± 14 years) (20 ± 64 years) (68 ± 73 years) (74 ± 90 years) Ratio (F-test)

Males (n) 18 64 29 15 Mean s.d. 1.261a 0.058 1.247a 0.081 1.203a 0.069 1.221a 0.081 P < 0.05{ (log 10) GM{ (mg=g protein) 18.24 17.66 15.96 16.63 (range; min-max) (14.92 ± 23.72) (13.03 ± 27.66) (11.02 ± 21.11) (10.81 ± 23.25) % < RNIx 5.6% 20.3% 34.5% 20.0% % < EARx 0.0% 0.0% 10.3% 6.7% % < LRNIx 0.0% 0.0% 0.0% 6.7% Females (n) 33 67 39 29 Mean s.d. 1.316x0.067* 1.277xy0.091* 1.252yz 0.082* 1.218z 0.094 P < 0.001 (log 10) GM{ (m g=g) protein) 20.70 18.92 17.86 16.52 (range; min-max) (15.83 ± 30.25) (10.40 ± 30.21) (12.80 ± 26.11) (8.87 ± 23.79) % < RNIx 0.0% 13.4% 20.5% 24.1% % < EARx 0.0% 4.5% 2.6% 10.3% % < LRNx 0.0% 1.5% 0.0% 10.3%

x: Reference nutrient intake (RNI) 15 mg=g protein intake, estimated average requirement (EAR) 13 mg=g protein intake, and lower reference nutrient intake (LRNI) 11 mg=g protein intake; (Department of Health, 1991). { Geometric mean; i.e. antilog 10 of log 10-mean. abc (xyz): Within males (abc) or females (xyz), means not sharing a common superscript are signi®cantly (P < 0.05) different; Post-Hoc ScheffeÂ test. {Though Variance Ratio is signi®cant, the Post-Hoc ScheffeÂ test indicates no difference between individual means. *Female mean is signi®cantly different (P < 0.05) from its corresponding male mean; Student's unpaired t-test. {{Means and s.d. of log 10-transformed data, geometric mean, ranges (min-max) and apparent percentage of subjects with intakes below RNIx, EARx and LRNx

intake but female intakes of vitamin B6 appeared to intake for either gender in any age group, whether vitamin decrease with age. B6 intakes were expressed as mg=dormg=g protein intake. Expressed as mg=d quite large percentages of all subjects (26 ± 69%) had intakes of vitamin B that were low in 6 Discussion comparison with the RNI (there is no LRNI as mg=d) but when expressed in terms of protein intake, only a small The method percentage of aged (>73 y) volunteers, of both genders, The production of a cyanide derivative of pyridoxal-5- had a tendency to fall below the LRNI (11 mg=g protein phosphate (PLP) has been applied successfully to human intake for vitamin B6 intake. plasma samples. The method is sensitive (requiring only 100 ml of plasma), rapid (elution time reduced to < 5 min), reproducible (2.8% coef®cient of variation) and Comparison of PLP concentrations with vitamin B6 intakes does not require speci®c anti-coagulants, offering an alter- There was no signi®cant statistical degree of association native to apo-enzyme methods for the routine determina- 6 between log10 PLP concentration and log10 vitamin B tion of vitamin B6 status. Plasma PLP by HPLC AL Bailey et al 453 The chromatographic separation of PLP and PA cyanide Intakes of vitamin B6 for the same volunteers, expressed derivatives (Figure 1), illustrates that pyridoxic acid (PA) in terms of protein intake (as recommended by the Depart- might also be determined by the method described here. ment of Health (1991)) however, do not indicate that the PMP and PM, although shown to produce ¯uorescent population groups studied are at the risk of vitamin B6 cyanide derivatives (Figure 1), were not found in plasma de®ciency, as indicated by PLP values. For all age groups samples at detectable levels ( < 2 mg=l). Other forms of the mean intake is above the RNI of 15 mg=g protein, vitamin B6 (PL, PN) did not appear to produce cyanide suggesting that on the whole these populations were con- derivatives under the conditions used, although PL has suming suf®cient B6. Only one adult (female) and four aged previously been demonstrated to react with cyanide under volunteers (3 females and 1 male) consumed less than the alkaline conditions to give a ¯uorescent product (Ohishi & LRNI, and thus were probably consuming inadequate Fukui, 1968). The measurement of PLP alone, however, amounts of vitamin B6. However, if vitamin B6 intakes was the purpose of the method presented, since it is known are expressed without reference to protein intake, quite to be the active coenzyme form of vitamin B6 in plasma. large percentages of all the groups studied appeared to be The detection limit of 2 mg=l (8 nM) could be lowered consuming inadequate amounts of vitamin B6 when com- considerably by use of a `state of the art' HPLC ¯uores- pared to the RNI. It should be noted that the RNI and LRNI cence detector. are based on very few studies (Department of Health, 1991). Although plasma PLP concentrations are believed Sample stability and quality control to re¯ect short-term (a few days) intake, in this study no Pooled plasma samples were found to be stable for at least relationship was found between vitamin B6 intake and 16 months, with respect to PLP concentrations, when stored plasma PLP for either gender in any age group, whether under liquid nitrogen. Borschel et al (1987) found plasma vitamin B6 intakes were expressed as mg=dormg=g protein PLP to be stable for up to 2 y when stored in the dark at intake. As part of a larger study the vitamin B6 intakes in 30C, but the radio-enzyme kit for PLP (Buhlmann the adolescent group were measured by direct analysis of Laboratories AG, Basel, Switzerland) states that EDTA foods (for concentrations of pyridoxamine, pyridoxal, and plasma remains stable at 20 C for only up to 4 weeks. If pyridoxine; the three major forms of vitamin B6 in food; pooled plasma samples can be aliquoted into small volumes Brubacher et al, 1986) and were compared with those and stored at 30C or less, it would appear that these determined by weighing and calculating from food table make a suitable in-house quality control material. For data (Southon et al, 1992). No signi®cant correlation was quality control between laboratories, samples would have found between vitamin B6 intakes obtained by direct to be transported under carefully controlled conditions. analysis and those obtained by calculation, and analysed Future investigations should include studies of the effects values were signi®cantly lower; possibly because nonspe- of storage and handling of samples in typical hospital ®eld ci®c microbiological methods for B6 content were used in study conditions. food table data (Paul & Southgate, 1978). This is still largely true of the latest edition of the food tables (Holland Plasma PLP concentrations in human population groups et al, 1991) since only a portion of the data has been Many studies have found reduced indices of vitamin B6 produced by more modern HPLC methods. Using directly status with age (Bitsch, 1993) but in the aged volunteers analysed vitamin B6 intakes no correlation was found here, (>73 y) males had signi®cantly increased mean PLP between intake and plasma PLP values in female adoles- concentrations when compared to younger age groups and cents, but there was a signi®cant correlation for boys there was no evidence of lower plasma PLP in other elderly (r 0.56, P < 0.05; Southon et al, 1994, Bailey et al, groups, when compared to younger groups. A possible 1997). Schuster et al (1984) found a positive correlation reason for this discrepancy might be that the elderly and between PLP and vitamin B6 intake from the previous day. aged population groups studied here were all free-living Vitamin B6 intakes when expressed as total mg=d were and, therefore excluded the portion of elderly of poorest higher for males than females, but the difference was only health (institutionalised or hospitalised), whereas some signi®cant for adult and aged populations. Since males other studies include, or concentrate, on elderly of poor consume more weight of food than females (Department health (Bitsch, 1993). The volunteers used by Rose et al of Health, 1991), one would expect mg=d vitamin B6 (1976) were also free living but the great diversity in intakes to be higher. When expressed per gram of protein, general `state of health' of elderly populations make it the present study shows females to consume more vitamin dif®cult to compare measurements between studies without B6 than males in all age groups except the aged. This more precise details of the health records of the volunteers. suggests that females consume a more vitamin B6 dense Clinical manifestations of vitamin B6 de®ciency are diet than males. The intake results have been expressed reported to be rare, since the vitamin is found widely both as mg=d and mg=d protein since the authors consider distributed in foods, and yet several groups within the that the recommendations of the Department of Health population have been described as having low biochemical (1991) should be considered, even if based on limited data. indices of vitamin B6 status. Examples include women That such large numbers of an apparently healthy during pregnancy and lactation (Vir et al, 1980; Guilarte, population (consuming apparently adequate amounts of 1993), adolescents (Korede & Ajayi, 1991), and the free- vitamin B6) should be at risk of vitamin B6 de®ciency is living, low-income elderly (Manore et al, 1989). The worrying. One explanation could be that intakes calculated adolescent, adult and free-living elderly populations stu- from food tables are an over estimation of actual intake, as died, using the modi®ed method described in this paper, was found by comparison of calculated and analysed intake show considerable proportions of the volunteers to be at for adolescents (Southon et al, 1992). If this is the case, and risk of de®ciency when PLP concentrations are compared the numbers of volunteers at risk of de®ciency is real, then to the currently used cut-off value for indication of vitamin the B6 nutrition of the population studied is a cause for B6 de®ciency (8.5 mg=l (34.4 nM); Rose et al, 1976). concern. However, a consideration of the way in which the Plasma PLP by HPLC AL Bailey et al 454 cut-off value for plasma PLP was established, throws doubt and Science and the Vitamin Forum. The authors would like to thank Mrs Zoe Piper for her invaluable analytical assistance. upon how appropriate the value is for judging vitamin B6 adequacy. The 34.4 nM cut-off was placed arbitrarily at the lower end of reference ranges for normal healthy male Abbreviations used subjects aged 18 to 90 y and was described by the authors PLP, pyridoxal-5-phosphate; PM, pyridoxamine; PL, pyr- (Rose et al, 1976) as classifying 30% of their unsupple- idoxal; PN, pyridoxine; PMP, pyridoxamine-5-phosphate; mented subjects as having marginal, or inadequate, B6 PA, 4-pyridoxic acid; HPLC, high performance liquid status. Thus, it would be expected that use of this cut-off chormatography; EC, electrochemical; TCA, trichloroace- value would indicate a similar percentage of subjects with tic acid; ODS, octadecyl silicate; MAFF, Ministry of inadequate PLP concentrations in other healthy popula- Agriculture, Fisheries and Food. tions. We suggest that if plasma PLP is to be considered as a routine measure of vitamin B6 nutrition, that more meaningful cut-off values for judgement of de®ciency needs to References be established. This 34.4 nM PLP cut-off value continues to Adams E (1969): Fluorometric determination of pyridoxal phosphate in be quoted and often acts as the basis for judgement of enzymes. Anal. Chem. 31, 118 ± 122. vitamin B6 adequacy (Gibson, 1990). Leklem (1990) sug- Bailey AL, Maisey S, Southon S, Wright AJA, Finglas PM & Fulcher FA gests a cut-off of 30 nM PLP, but this still de®nes large (1997): Relationships between micronutrient intake and biochemical percentages of the population as at risk of vitamin B indicators of nutrient adequacy in a free living elderly UK population. 6 Brit. J. Nutr. 77, 225 ± 242. de®ciency. If plasma PLP is to be of any use as a measure Bender DA (1992): Pharmacological uses of vitamin B2.InNutritional of vitamin B6 adequacy, either alone or together with Biochemistry of the Vitamins, Bender DA. Cambridge: University Press, another index such as an enzyme activation assay, further pp 255 ± 258. work must be done to establish a meaningful cut-off value Bender DA (1993): Lack of concordance between two biochemical indices of vitamin B6 nutritional status. Proceed. Nutr. Sci. 52, 315. against which PLP results for population groups can be Bender DA (1994): Novel functions of vitamin B-6. Proceed. Nutr. Soc. compared. 53, 625 ± 630. [It must be noted that the reference ranges given by Bitsch R (1993): Vitamin B6. Inter. J. Vit. Nutr. Res. 63, 278 ± 282. Rosalind Gibson (1990) are erroneous in that an incorrect Borschel MW, Kirksey A & Hamaker BR (1987): A micromethod for molecular weight appears to have been used in converting determination of plasma pyridoxal phosphate and its assessment of storage stability of the vitamer. J. Pediatr. Gastroent. Nutr. 6, the values of Rose et al (1976): 8.5 mg=l is equivalent to 409 ± 413. 34.4 nM and not 50.8 nM as quoted by Gibson (1990)]. Brubacher G, Muller-Mulot W, Southgate DAT et al (1986): Vitamin B6 in Foodstuffs: HPLC method. In: Methods for the Determination of Vitamins in food recommended by COST 91., Brubacher G, Muller- Mulot W & Southgate DAT (eds), London: Elsevier Applied Science Conclusion Publishers, pp 129 ± 140. Coburn SP, Lewis DLN, Fink WJ, Mahuren JD, Schaltenbrand WE & A modi®ed HPLC ± ¯uorescence technique has been Costill DL (1988): Human vitamin B-6 pools estimated through muscle applied to the analysis of pyridoxal-5-phosphate (PLP), biopsies. Am. J. Clin. Nutr. 48, 291 ± 294. Coburn SP, Ziegler PJ, Costill DL, Mahuren JD, Fink WJ, Schaltenbrand the predominant active form of vitamin B6 in human WE, Pauly TA, Pearson DR, Conn PS & Guilarte TR (1991): Response plasma. The modi®ed method gives good recoveries of of vitamin B-6 content of muscle to changes in vitamin B-6 intake in added PLP and small variability, requiring only 100 mlof men. Am. J. Clin. Nutr. 53, 1436 ± 1442. plasma for analysis. The chromatographic elution time of Department of Health (1991): Vitamin B6. In: Report on health and social less than 5 min enables a high sample throughput (20 ± 30 subjects: Dietary Reference Values for Food Energy and Nutrients for the United Kingdom, Department of Health. London: HMSO, pp 102 ± 105. samples=d) and is highly suited to automatic sample injec- Gibson RS (1990): Assessment of vitamin B-6 status. In Principles of tion. Nutritional Assessment, Gibson RS (ed). New York: Oxford University Plasma PLP concentrations are given for adolescent, Press, pp 445 ± 457. adult, elderly and aged human population groups. Percen- Guilarte TR (1993): Vitamin B6 and cognitive development: Recent research ®ndings from human and animal studies. Nutr. Revs. 51, tages of subjects with PLP levels below the 34.4 nM cut-off 193 ± 198. would appear to indicate that large proportions of the Holland B, Welch AA, Unwin ID, Buss DH, Paul AA, Southgate DAT apparently healthy population within this community are (1991): McCance and Widdowson's The Composition of Foods. 5th edition. Cambridge, UK: Royal Society of Chemistry. at risk of vitamin B6 de®ciency. This was not re¯ected by vitamin B intakes calculated from food tables, but these Kleijnen J & Knipschild P (1991): Niacin and vitamin B6 mental 6 functioning: A review of controlled trials in humans. Biol. Psych. 29, are possibly an over-estimation of actual intake. However, 931 ± 941. the cut-off value for plasma PLP was placed arbitrarily and Kleijen J, Ter Riet G & Knipschild P (1991): Vitamin B6 in the treatment evidence to date indicates it is set too high. of premenstrual syndrome. Brit. J. Obstet. Gynaecol. 98, 329. Korede O & Ajayi OA (1991): Plasma vitamin B6 concentrations in Nigerian adolescents. Euro. J. Clin. Nutr. 45, 111 ± 115. Contributors Leklem JE (1990): Vitamin B6: A status report. J. Nutr. 120, 1503 ± 1507. All authors were involved in initiation, execution, inter- Li T-K & Lumeng L (1980): Plasma PLP as an indicator of nutrition pretation of results and preparation of the ®nal typescript status: Relationship to tissue vitamin B-6 content and hepatic metabo- and subsequent revision. AL Bailey also performed the lism. In: Methods in vitamin B6 nutrition, Leklem JE & Reynolds RD (eds). London: Plenum Press, pp 289 ± 296. method of validation, HPLC analysis of plasma samples Manore MM, Vaugham LA, Carroll SS & Leklem JE (1989): Plasma and preparation of the manuscript. AJA Wright also super- 0 pyridoxal 5 -phosphate concentration and dietary vitamin B6 intake in vised blood sample handling and performed the statistical free living, low income elderly people. Am. J. Clin. Nutr. 50, 339 ± 345. analysis of results. S Southon was the principle investigator Murray M & Pizzorno J (1990a): Atherosclerosis. In: Encyclopedia of and oversaw all aspects of the study. Natural Medicine, Murray M & Pizzorno J. (eds), London: Macdonald Optima, pp 156 ± 170. Murray M & Pizzorno, J (1990b): Premenstrual Syndrome. In Encyclope- Acknowledgements ÐThis work was supported by the UK Ministry of dia of Natural Medicine, Murray M & Pizzorno J. (eds), London: Agriculture, Fisheries and Food (MAFF), the UK Department of Education Macdonald Optima, pp 470 ± 479 & 601. Plasma PLP by HPLC AL Bailey et al 455 Naoi M, Ichinose H, Takahashi T & Nagatsu N (1988): Sensitive assay for Southon S, Wright AJA, Finglas PM, Bailey AL, Loughridge JM & the determination of pyridoxal-5-phosphate in enzymes using high Walker AD (1994): Dietary intake and micronutrient status of adoles- performance liquid chromatography after derivatization with cyanide. cents: Effects of vitamin and trace element supplementation on indices J. Chrom. 434, 209 ± 214. of status and performance in tests of verbal and non-verbal intelligence. Ohishi N & Fukui S (1968): Further study on the reactions of pyridoxal Br. J. Nutr. 71, 897 ± 918. and Pyridoxal-50-phosphate with cyanide. Archiv. Biochem. Biophys. Stewart JW, Harrison W, Quitkin F & Baker H (1984): Low B6 levels in 128, 606 ± 610. depressed outpatients. Biol. Psych. 19, 613 ± 616. Paul AA & Southgate DAT (1978): In McCance & Widdowson's The Tamura Z & Takanaski S (1970): Fluorimetric determination of Composition of Foods, 4th Ediction. London: Her Majesty's Stationery pyridoxal and pyridoxal-5-phosphate in biological materials by Of®ce. reaction with cyanide. In: Methods of Enzymology, Vol 18, Rall LC & Meydani SN (1993): Vitamin B6 and immune competence. McCormick DB and Wright LD (eds), New York: Academic Press, Nutr. Revs. 51, 217 ± 225. pp 471 ± 475. Rose CS, Gyorgy P, Butler MS, Andres R, Norris AH, Shock NW, Tobin J, Ubbink JB (1994): Vitamin nutritional status and homocysteine: An Brin M & Spiegel H (1976): Age difference in vitamin B6 status of 617 atherogenic risk factor. Nutr. Revs. 53, 383 ± 387. men. Am. J. Clin. Nutr. 29, 847 ± 853. Vanderslice JT, Brownlee SG, Cortissoz ME & Maire CE (1985): Vitamin Sandyk R & Pardeshi R (1990): Pyridoxine improves drug- B6 analysis: sample preparation, extraction procedures, and chromato- induced parkinsonism in a schizophrenic patient. Int. J. Neurosci. 52, graphic separations. In Modern Chromatographic Analysis of Vitamins. 225 ± 232, Eds, De Leenheer AP, Lambert WE & Rayter MGM (eds), New York: Sauberlich HE (1984): Newer laboratory methods for assessing nutriture of Marce Dekker, pp 435 ± 476. selected B-complex vitamins. Ann. Rev. Nutr. 4, 377 ± 407. Verhoef P & Stampfer MJ (1995): Prospective studies of homocysteine Selhub J, Jacques PF, Wilson PWF, Rush D & Rosenberg IH (1993): and cardiovascular disease. Nutr. Rev. 53, 283 ± 288. Vitamin status and intake as primary determinants of homocysteinemia Vermaak WJH, Barnard HC, Potgieter GM & Theron Hdu T (1987): in an elderly population. J. Am. Med. Assoc. 270, 2693 ± 2698. Vitamin B6 and coronary heart disease. Epidemiological observations Schuster K, Bailey LB, Cerda JJ & Gregory III JF (1984): Urinary 4- and case studies. Atherosclerosis 63, 235 ± 238. pyridoxal acid excretion in 24-hour verus random urine samples as a Vir SC, Love AHG & Thompson W (1980): Vitamin B6 status during measurement of vitamin B6 status in humans. Am. J. Clin. Nutr. 39, pregnancy. Inter. J. Vit. Nutr. Res. 50, 403 ± 411. 466 ± 469. Wright AJA, Southon S, Bailey AL, Finglas PM, Maisey S & Fulcher RA Southon S, Wright AJA, Finglas PM, Bailey AL & Belsten JL (1992): (1995): Nutrient intake and biochemical status of non-institutionalized Micronutrient intake and psychological performance of school children: elderly subjects in Norwich: comparison with younger adults and consideration of the value of calculated nutrient intake for the assess- adolescents from the same general community. Brit. J. Nutr. 74, ment of micronutrient status in children. Proc. Nutr. Soc. 51, 315 ± 324. 453 ± 475.